Atten4Vis/ConditionalDETR

Issues

• coco indicator is 0

  #34 opened 2 years ago by 304999183
  1

• About the multi-head attention

  #25 opened 2 years ago by LiewFeng
  3

• The diagnoal matrix meaning?

  #16 opened 3 years ago by JosonChan1998
  4

• Issues about Positional Embedding and Reference Point

  #32 opened 2 years ago by tae-mo
  1

• HELP

  #33 opened 2 years ago by 304999183
  1

• log and model of Group C-DETR

  #36 opened a year ago by LiewFeng
  0

• How to download CDETR pretrained weights?

  #35 opened 2 years ago by ppeterpp
  2

• the parameters are only 43196001, instead of 43524961

  #14 opened 3 years ago by Cohesion97
  3

• Visualization code of Figure 1 in paper.

  #5 opened 3 years ago by MaureenZOU
  6

• questions about provided conditional detr model

  #31 opened 2 years ago by xz-123-new
  3

• Add ConditionalDETR to HuggingFace Transformers

  #21 opened 2 years ago by NielsRogge
  3

• groupcondetr pretrain

  #30 opened 2 years ago by lijain
  1

• How to add Group DETR in DINO-Deformable-DETR

  #29 opened 2 years ago by xiaoruiai
  4

• What's the difference between the Group DETR and the DETRs with Hybrid Matching?

  #28 opened 2 years ago by rockywind
  1

• some question about lr

  #27 opened 2 years ago by LYMDLUT
  1

• cannot import name '_LinearWithBias' from 'torch.nn.modules.linear'

  #18 opened 3 years ago by cxq1
  5

• Training Error assert (boxes1[:, 2:] >= boxes1[:, :2]).all()

  #17 opened 3 years ago by Kyfafyd
  1

• If there's a bug? target_classes_onehot.scatter_(2, target_classes.unsqueeze(-1), 1)

  #24 opened 3 years ago by heran
  0

• Hello, may I ask how to visualize the spatial attention diagram in the paper? Where is the corresponding code?

  #20 opened 3 years ago by wulele2
  6

• Can you provide the checkpoints of ConditionalDETR-R50 and -R101 trained with 108 epochs ?

  #15 opened 3 years ago by truetone2022
  6

• what's the main purpose of concat (pos_y, pos_x) instead of (pos_x, pos_y)?

  #19 opened 3 years ago by kuixu
  1

• Encoder ends with two LayerNorm layers

  #11 opened 3 years ago by joakimjohnander
  1

• How to use resume correctly?

  #12 opened 3 years ago by xziyh
  6

• How extract precision, recall and f1-score metrics

  #13 opened 3 years ago by jackonealll
  1

• out of memory

  #10 opened 3 years ago by xziyh
  3

• Default PostProcess behavior is changed compared with DETR

  #8 opened 3 years ago by YuyaoXiaoCS
  1

• small typo in readme

  #9 opened 3 years ago by XiaoyuShi97
  0

• How to finetune ConditionalDETR on custom dataset?

  #7 opened 3 years ago by ducvuuit
  1

• How to visualize the attention map of the cross-attention

  #4 opened 3 years ago by GWwangshuo
  1

• The code of validate the model performance (AP，AP50)

  #6 opened 3 years ago by Huzhen757
  1

• Some question about your code and Cross-attention module in your paper

  #3 opened 3 years ago by Huzhen757
  5

• Some questions about design choices

  #2 opened 3 years ago by weii41392
  1

• What is the speed of the model Conditional DETR-DC5-R50?

  #1 opened 3 years ago by tangjiuqi097
  2